Different from analog signals and digital signals, the signal transmission among nerve cells is known to be realized by electric signals that resemble an impulse train corresponding to pulse frequency modulation. The Hodgkin Huxley equations for describing the ignition mechanism of nerve cells are also well known. According to the Hodgkin Huxley equation, the nerve cell membrane can be expressed by an electric equivalent circuit of a certain type and the nerve activity can be expressed by expressing the conductance relative to ions by means of a function of voltage and time.
Additionally, an artificial nerve circuit shown in Patent Document 1 is known as an equivalent circuit simulating the ignition phenomenon.
Massive research efforts have been and being paid for the neural network in the field of artificial intelligence and automatic control. The sigmoid function (f(x)=1/(1+exp(−x)) or the like is employed for each unit of the network (see Non-Patent Document 1).
Patent Document 1: Japanese Patent Application Kokai Publication No. Hei 07-168901
Non-Patent Document 1: D. E. Rumelhart, G. E. Hinton & R. J. Williams, “Learning Internal Representations by Back-Propagation Errors”, Nature, 323, 533-536, Oct. (1986)